MONDAY, 15 JUNE 2020
Below is a transcript of the latest episode of the BlueSci Podcast, which we have provided to help our podcast reach a wider audience. The script has been transcribed using an audio-to-text converter (otter.ai) and edited by our hosts. If you spot any errors please do let us know! Please do subscribe, follow, like, and review our podcast to show us your support and give us feedback, whether you listen or read the episode.
You can find the podcast on:Meet the graduate students behind the COVID-19 educational webapp, featuring Daniel Muthukrishna and Nick Taylor
AnchorFM : anchor.fm/bluesci-podcast
Apple Podcasts: podcasts.apple.com/podcast/id1358079746
Google Podcasts: podcasts.google.com?feed=aHR0cHM6Ly9hbmNob3IuZm0vcy8zZWRmMTdjL3BvZGNhc3QvcnNz
AnchorFM : anchor.fm/bluesci-podcast
Apple Podcasts: podcasts.apple.com/podcast/id1358079746
Google Podcasts: podcasts.google.com?feed=aHR0cHM6Ly9hbmNob3IuZm0vcy8zZWRmMTdjL3BvZGNhc3QvcnNz
Released 15th June, 2020
BlueSci Podcast 0:05
Welcome to the Bluesci podcast, brought to you by Cambridge University Science Magazine. I'm Ruby and I'm Simone. Every two weeks we speak to local researchers, university staff and students and anyone who works in science to learn about their research and activities, hear about the work that they do, and uncover what goes on behind the scenes. If you want to get in touch with a question, suggestion or just want to be featured on the podcast, just drop us a tweet. Our handle is @bluescipod and you can also email us at email@example.com
This week, we're talking to two PhD students, Nick Taylor from the Department of Plant sciences, and Daniel Muthukrishna from the Institute of Astronomy. Together, with Dr. Cerian Webb (who's a postdoc in the theoretical and computational epidemiology group in the plant sciences department), they developed an online educational tool called the lowhighcovid web app, so that anyone that is interested in learning more about infectious disease modelling and some of the scientific concepts behind the news that you might be reading about the Coronavirus. They have a data feed which changes in real time where you can see the data regarding Coronavirus spread in different countries and you can compare different countries. They also have a bunch of educational videos on all these concepts like the R values and how modelling works. And they also have a space where you can kind of make a fictional country, decide what your population is, what the infection rate is, and then apply different measures like social distancing, or complete lock downs and then see how the spread of the virus changes. So it's an interactive way to learn more about how the Coronavirus and generally any infectious disease can spread in society and understand how the government policies behind that might be working. And it's really a really great resource to for anyone. You don't have to be a scientist to understand it. So you can go onto their website by clicking the link in the description. And the reason we invited them onto the podcast was to hear a bit about how they came up with the idea of making this educational tool. So they're not, you know, experts in the subject themselves, but how they came together as students want to learn more about the subject, and to hear about their perspectives, and how we can use these types of communication tools to communicate science in a more effective way, especially when we are living through a situation where a lot of people that are perhaps not that familiar with these concepts now want to understand a little bit better.
The BlueSci podcast is sponsored by Greiner Bio-One who supply laboratory diagnostic and medical products to research institutions, higher education, the NHS and others across the UK. For details of the full product range, visit www.gbo.com.
Welcome Nick and Dan. This is our first episode recording with two interviews at once, so thanks for joining us and hopefully wecoordinate around this. So you've been involved in developing this low high COVID app. And we're wondering whether you could just sort of briefly give us an overview of what it's all about.
Yeah. So the app was kind of designed just as a project that wekind of sketched up in Claire bar one evening. But basically, the idea is, is to try and communicate some of the scientific modelling that's gone on, to kind of the principles behind what drove the decisions that were made back in March to start a lockdown in the UK, and also just kind of further explore how we thought the epidemic might progress. And then Dan sort of came in and added a data aspect. And that was to sort of allow you to compare different different countries and how they approached tackling the pandemic
At the start in like, March, the UK kind of switched its tune from a kind of hands off or at least a relatively hands off approach from this herd immunity approach to something that changed into a lockdown approach and so we wanted to explore what these different approaches how much impact they really make. And Nick uses these things called SIR models before and done a lot of that stuff in his epidemiology work. And so he had a lot of experience there and it's not it's not too much of a stretch to go from plants to humans.
Yeah, so I'm based in plant sciences and modelling plant epidemics, but the sorts of models to apply to human epidemics are very similar, at least the more basic ones. So, I, I sort of adapted some of those to work for the Coronavirus pandemic, and then using quite a few of the parameter values from the Imperial paper that was published and that was sort of informing a lot of Government policy in in March or so. So it was using what are called compartmental models. So you sort of split the population up according to disease status. So sort of partitioning into people who are healthy or infected, or, and I guess you have different levels of infection. So, depending on the severity of the infection, you can progress to being hospitalised or you can recover. And then if you're hospitalised and you can progress to being in a critical state and then either recovering or dying.
From that, was that the kind of motivation behind the app because you said it was kind of like a kind of spontaneous idea that you had together because I that was gonna be one of our questions like, how did you? Obviously you were both based in different departments, and how did you come together to make this specific project?
I guess, yeah. So Dan, and I were friends from from Clare College. I guess part of it was just an interest in how everything was going to progress, so we had a personal interest in how the epidemic was likely to affect us and our friends. And that developed into putting more and more time into this project to try and understand the pandemic to greater levels of detail. And then one of the postdocs from my lab in plant sciences, Dr. Cerian Webb, thought it'd be great to push this as a sort of more educational platform approach. So once we'd developed this model to actually sort of get it online and try and push it as an educational tool.
Yeah. So I think I think it was about like, in March, when University kind of closed down, the government, the country kind of closed down in some sense, and that was all that anyone could really think about was Coronavirus. And so for us, it was kind of, you know, it's kind of difficult to work it out own on your own studies with so much going on in the world. And so instead of just thinking about it...now I know that I would like every day look at the numbers, how many? How many? How many new deaths how many new cases have happened today? And so that, you know, every day I'd be looking at Worldometer to see what what was changing. For me, that inspired me, and I wanted to be able to compare countries. So one thing that became clear was that you know, in March, you know, at the start of March, the UK had zero deaths, right. And then by the end of March, we had over 2000, by the end of April, we had 20,000. So what became clear was that we didn't really understand exponential growth very well. It was quite, you know, it's quite difficult for people to get their heads around this. And so, you know, in March, we talked about how bad Italy was, for example, as this you know, this place of like, what is going on there? The Coronavirus is really hitting there badly. And what became more clear was that it you know, this isn't a problem that's specifically Italy, its just that Italy got there first, and within two weeks, we would be there as well. And so I wanted to be able to see this I just want to be able to like, you know, can I, can we create it? So there are three kind of parts to this app. There's the modelling part of it, which Nick really worked on quite well. And then there's the data part of it, where we can kind of just analyse different countries and compare what's going on, which I worked on quite a lot as well as the background knowledge which Cerian worked on quite a lot. And so with the data part of it, I wanted to be able to see, you know, can we actually model exponentials onto these and actually see, well, how far behind are we each other, each country? For me, it was actually it started off as being something I wanted to look at. And so to me, and Nick, we kind of wanted to understand a little bit more, and then we became, it became a little bit more of like a public service. So, you know, I think it's gonna help a lot of people to try to understand the Coronavirus a little bit better, what kind of government measures are going on? And you know, how the data is changing with time. And so we made this into a nice user friendly app, and we thought that this might be actually useful to a lot of people.
Yeah. Have you had any kind of response from have people gotten in touch with you? Or do you know if it's being used as an educational tool maybe in the classroom?
Yeah, I think one thing we can track is like what kinds of people are using it. They're being used in places you'd really expect. So we have South America, obviously a lot of England and just like Peru and Brazil and all these other countries, which somehow found the app as well. So it is being used quite widely, not quite sure for what purposes but I do know that a lot of people had. You know, as we released a lot of feature requests like can you add this, can you add this kind of like, you know, I have a few friends around the world. Well, can you add our country to this as well? If we compare this and people were very interested in, you know. Coronavirus, had really hit a few countries quite badly, but other countries have started to get there. So they wanted to kind of see where they were on this scale. And so they asked that, abd I know that one of my one of my collaborators in the US used it in his time series analysis class as an example of how to model these time series data. This is an example of what's going on.
Sounds like it's massively successful, and it must feel awesome to see people all over the world using it. And I think what really struck me when I was having a play with it as well, I kind of realised that these kind of apps aren't sort of that common or that easily available. Was that kind of a motivation as well? As a condensed source of information because there's so many numbers flying around. And it's very simple. It's very user friendly. And it's very sort of graphical, which I think even if you're not scientifically minded, you can engage with. Was that a motivation behind it as well, or did it sort of evolve into that?
Yeah, it kind of snowballed a bit. So I sort of got interested in this modelling, probably mid March or so. I was kind of feeling like I'm doing a PhD in mathematical epidemiology, but at the moment, all my projects focused on plants, but it felt a bit strange to ignore the sort of biggest pandemic to to humans for sort of 100 years. So that's when I thought actually, I want to get into modelling some of this. And my supervisor was really supportive of that. And that allowed me to spend some of that time building up this app. And then as I saw, sort of more of these models being published and a few of these sort of web interfaces coming out online, I saw a lot of interfaces where you've got so many parameters being thrown at you. And to me, I felt the only people that would understand that were other disease modellers, who could probably and probably had already written their own models. So I felt I wanted to try and create something. I guess it's difficult to know what level to pitch it out. But I was trying to make something that was, was hopefully slightly more accessible. And I think, each update to the app, I've tried to make it more more accessible and intuitive where I can, but again, before I'd done this, I hadn't done much sort of web programming. So that's been another kind of thing to explore, I guess. I think it's, you know, it's not the perfect polished product yet. But it's, it's been a really interesting kind of project for me, and I'm sure Dan would say the same.
I think having, making, it user friendly was quite important. But we started off just making this for ourselves kin d of on just, you know, a Python backpack and just doing it. And we thought, you know, it's, we got to build each part of it independently on the app and then put them together. And I think, you know, it's a really good way to just let average people try and play with, like, in an interactive way, because, you know, the, the real analysis that showed what was going on was this, this paper that came out of Imperial, Imperial College about, you know, what the government should be doing, and that was very intractible for most people to really understand. And so putting this into a model where people can actually play with parameters and play with, you know, what country they want to analyse, they play with what the starting population is, what kind of what rate of susceptibility is what the death rate is, and actually put those numbers in themselves and just play with what kind of scenario is happening, you know, what is gonna happen when you lock down the population versus social distancing population. And we're also really curious about, you know, we call it low high, because what became clear was that this virus is obviously very, very deadly, but it was hitting high risk people. So people with pre existing conditions a lot more significantly than it was before with conditions that want to serious and so back, when you know, the word herd immunity was being thrown around quite a lot. It became clear that you know, this would only really work if the people who were getting sick were low risk. And so we wanted to consider, well, you could lock down everyone but what would happen if you just locked down a higher set of the population that lowest set of the population. And we we let you play with the web app, you can actually play with what happens when you do each of those scenarios. And I thought getting that information out to the public and letting public actually play with these and see what's going on. I think Nick and I both thought that was a pretty important thing to do.
Yeah, and I guess it's the kind of thing where, you know, when you want to understand what's going on at the moment, and realise that there isn't really a way to do that in the way that you want, you end up doing it yourself, and then automatically helps other people that have the same question as you. And I guess, do you think that, you know, speaking more broadly, that's something that we struggle, not struggle with, but could do better as a society to, even though we don't have that kind of scientific literacy in our day to day of, of knowing about, you know, how these government policies suddenly use this paper and then something goes through and we have no idea you know, how that works, or do you think that more can be done from you know, a scientist's perspective to kind of talk about these things more and do more to, you know, talk to members of the public in general about how these concepts like play about in our day to day lives, because obviously now we know how the Coronavirus affects us, you know, quite a lot because everyone's thinking about it. But there's probably a lot of other aspects of our lives that we don't you know, know about in that sense that we could all benefit from. I don't know, what are your thoughts on that?
Yeah, definitely. I think that the public engagement aspect is a really important one. And I think it's one that is sometimes overlooked. Just communicating this, you know, this great work that that is done by researchers across, you know, within the university, but across the country, and that doesn't always, if it does reach the public, sometimes it's through a headline and the headline is sort of lifted from a, you know, from one sentence within the paper that isn't necessarily representative of the science. So yeah, I think trying to facilitate that, that conversation between, you know, research and and public, it is a really interesting one.
I think we, as scientists have a duty to, you know, we're experts in understanding data, a lot of us and so an average person might struggle to actually interpret what's going on, because they haven't dealt with them on a daily basis. And so I think we have a duty to be able to present that and explain the ways to community. And so I think this is a real opportunity to do that.
I think it has been really interesting, actually, throughout the pandemic, how how important in this case is to communicate pretty technical, scientific detail to the public. And, you know, you see some people up, you know, on the news who do that really effectively, and you see some that maybe don't do it so effectively, and it's not an easy, easy thing to do. But I think that's been really interesting for me just to see you know, which approaches are kind of capturing people's interest and really conveying that message and which ones are failing?
Yeah. I mean, you know,to have anyone to believe that this is going to be a huge pandemic, when you only have, you know, less than 10 deaths in the in the UK. That's pretty hard to believe, you know, but the models, if the model say that's what it is, scientists need to be able to explain that to people. And I think the thing is very difficult for people to really accept that, you know, when the numbers are so low, that they're suddenly going to change. And this is true for this current virus and all exponential growth, including climate change and all sorts of other issues.
Yeah, and that's been a really interesting one is that the numbers that are quoted a very often the deaths because that's the, you know, the key impact that this virus has had. But that's not a particularly useful metric, in terms of predicting how it's going to progress in the future. Because the death because it kind of takes, you know, approximately four weeks or a month from getting the infection to progressing to, to a serious infection and then dying. And that means that the death figures relate to stuff that happened a month ago. So it's not a very good predictor of, of how we're dealing with the virus right now. So that's why even after lockdown happening, sort of mid to late March, we had, you know, a huge peak in terms of infections. And then slightly after that we had a peak and deaths. And April was was where we were hit really hard in terms of death figures.
Yeah, I think one thing that's interesting is that, you know, the public really got behind this message that you know, it was quite obvious what was going on, and it became more and more obvious with time that you know, this is this is happening, this is what's changing, and that's partially because of, you know, good scientific communication that was happening in some parts. But also partially because it was, you know, this, this kind of exponential growth is happening on a daily timescale, you can kind of see. But you know, when you've got other issues where exponential growth is happening on a yearly timescale, people are less likely to believe that. And so it really takes scientists to come out there and try and really get this message across rather than just telling people try and get them to understand these kind of apps can really help people play with the models and actually see, well, what is going on. Because, you know, in other issues in the world, you know, we're expected to believe these things, the average person and that's, that's not difficult. That's not easy to do if you don't really understand the science. And I think these interactive apps really allow people to try and understand the science a little better. So we can do these with other global problems such as climate change and other things. I think that's, that's really great as well. And it's a good way to get the message out there in ways that people can understand because believing scientists isn't enough. I mean, it's becoming harder and harder to do. But if you have the model that you as a layperson and a non expert can use I think that's really great.
Yeah, we tried to keep documentation of how the model works. And at times that gets quite technical. But it is interesting to try and list those assumptions that you're making, because no model is going to be perfect. And I think you have to be very transparent about what assumptions are you feeding into a model. And one of the things that I think has sometimes failed to be communicated is, is the level of uncertainty in in these things. So with something like exponential growth, where the number of cases can skyrocket, I remember seeing paper that had confidence intervals for the number of infections. It was a sort of projection for the number of infections in a month time, and it said they're going to be 20,000 infections, but the lower bound was something like four or 500 and the upper bound was 10 times as many. And so for some of these things, the exact time that these processes will happen is difficult to predict. But as long as you're kind of clear about the range of values that are feasible, I think that's another really important thing to communicate is that that these models are imperfect. But that doesn't make them not useful. It just means that you need to kind of understand that there is going to be some kind of confidence interval around whatever prediction you get.
Yeah, absolutely. I think that's the curse of being a scientist in a way because we're sort of trained never to say that our results are, or whatever we're sharing, is 100% true. And I think sometimes people who aren't in science as much take that as doubt. And especially with science communication, I find that it is kind of a big hurdle to get over to try and get took the stand. But you know, you're 95% sure. But you can't ever be totally sure. But yeah, back back to what you were sort of talking about how apps like this in the future could potentially be applied to other situations, other sort of outbreaks, climate change and things like that. And if that's something you'd both be planning for, or if not, is that other sort of some examples that you can really think of? And then you mentioned climate change, but is there anything else?
I mean,our expertise is actually, well, you know, we're not biologists, we're not experts in this, we can look at that data. So I mean, I'm an astrophysicist. So, it's not it's not something I directly use. But I think if we can try and understand the models, which we did do, like, you know, we both read the, the papers that were going out, Nick actually works on this stuff on a daily basis. And you know, if we can try and understand something and convey that I think that's great. I don't know if Nick has any more plans. But, you know, when something else, I think the real motivation here was that, you know, this was all we could really think about. So such a massive part of our lives. So yeah,
Yeah, I mean, that's why I found the whole process quite rewarding, because I think, is a quite, quite cool project to work on. So it's certainly something that I'll use in my research. So it'll probably be similar things at some point for plant epidemics. But potentially other things. As well, I think it is quite interesting to, I think a lot of modelling a lot of the like, the interest around a model is actually just playing with it and getting a sense of what does it what sort of predictions that it generates, how does it work, which parameters are important and a lot of that can get lost in a paper where you just publish a few plots that have come out with that model. And you don't necessarily have a feel for exactly how it works. What are the important factors? I think it's a lot easier to engage with that if you can really see it and play with it. So that was the kind of interactive, uh, the sort of motivation for the interactive element.
You said that in terms of working with each other, obviously, you both come from very different backgrounds. And if you were working with this kind of modelling more on day to day basis, then you have the more like data side or the data, data processing aspect of it. Did you find that working together? Although like you said, you worked independently at the beginning does that did you were able to like learn from each other and learn about the kind of things that you both work on from this collaboration? And also from the third person that worked on the kind of videos explaining those concepts. Was that kind of a way to have you noticed that those skills have improved?
Yeah, I found it really useful, actually. So Dan's a final year PhD student. So I found it useful, sort of collaborating with him on some of the code because his coding is a lot neater than mine can be. So, and he had more experience of building these apps. So he'd made one before. And this was the first major one that I'd put together. And then Cerian from my lab does a lot of teaching at university. And so she she was really helpful in putting together sort of educational aspects and trying to communicate that message slightly more clearly than I think it would have been, would have been initially. So yeah, I think that collaboration was I really enjoyed it. Definitely.
Yeah. My background is software engineering where I do a lot of the stuff before before I became an astrophysicist. And so that that was useful for me but I had never ever really done any of these of these SIR models or anything like that. So I definitely learned a lot from Nick about how to actually do these models, you know, what the standard differential equations you need to plug in. And so that was, that was nice. It was definitely a good learning experience from these very different fields.
So you're both, PhD students and and both like in different areas. So could you tell us just quickly, what do you work on in your sort of day job, outside of coronavirus? Yeah.
Yeah, so I, so I 'm doing a PhD at the Institute of Astronomy. So a mainly work on machine learning and artificial intelligence applied to big data problems in astrophysics. And so I deal with these particular things called supernova, which is exploding stars at the end of their lives. I try to understand that a little bit more. These are so basically, you know this, these are time series events where you get nothing and then all of a sudden, you get a massive brightness in the sky in through a telescope where a star explodes. And from this change in brightness, we can understand these objects a little bit better. And so that's what my research is I'm trying to understand these objects that are trying to model them try to classify them using machine learning. Yeah. So it's very different to this. But you know, it's it's still still time series data in some sense.
Yeah, and my my PhD work is in, in plant epidemiology. So the problem I'm working on at the moment is to do with how pathogens evolve resistance to the sorts of control measures that we have. So for diseases of wheat that I've been working on, the main one is called septoria. And that's the major disease of sort of winter wheat that's grown in the UK, in Britain in Europe, and the control measures that we have using sort of disease resistance crop varieties and using chemicals. So fungicides and to try and control the pathogen. But if you keep deploying the same methods of controls and the pathogen involves resistance, so I'm trying to use these compartmental models where you have, you're trying to track the infection and how it progresses through a growing season. And within these models, you kind of have competition between the different strains of pathogen and you're trying to minimise this selection pressure that's leading to eventually pathogen strain that you just can't control very effectively. So that's, that's what my sort of day to day PhD work is.
So interesting about like, different backgrounds, and obviously, your day to day work, thinking about different questions, but actually, in terms of the skills that you can apply, you can apply into so many different ones and it just shows how, you know, like that kind of interdisciplinary perspective is so important and can be so helpful to understanding these kinds of issues. And hopefully, that's something that people will reflect on during this time and going forward will be something that people are more likely to do as well.
Yeah, I was just gonna say, I think a lot of science, at least a PhD level is quite individual. And I think there's probably a lot to be said for slightly more collaborative efforts, or at least trying to try to learn from other people's skill sets and approaches to to the same problem. And I found that really valuable.
Yeah, I think that's right. I think sometimes there's a tendency for PhD students to feel a bit isolated. You're in your own little bubble. But I guess it just shows that you can, you can come together and get projects that are really multidisciplinary and, you know, you'd never think of astrophysics and plant sciences coming together. It's awesome. It's really cool. Thank you so much for chatting to us today. And I'm sure all of our listeners would be really keen to try out your app.
Thanks so much for having us.
We really hoped you enjoyed this episode chatting to Nick and Dan about the app that they developed with Dr. Cerian Webb about Coronavirus, and modelling Coronavirus. It's called low high COVID and I highly recommend checking out, it's really, really interesting. You can play with all sorts of parameters and control measures. And it really really helps you understand COVID a little bit better, as well as getting more comfortable with modelling and what modelling actually is. I found it super helpful. And yeah, if you enjoyed this episode, please like and subscribe. And you can contact us at @bluescipod on Twitter, and you can also email us at podcast at firstname.lastname@example.org. Bye for now and see you next time.